Alert button

Sponge Examples: Energy-Latency Attacks on Neural Networks

Jun 05, 2020
Ilia Shumailov, Yiren Zhao, Daniel Bates, Nicolas Papernot, Robert Mullins, Ross Anderson

Figure 1 for Sponge Examples: Energy-Latency Attacks on Neural Networks
Figure 2 for Sponge Examples: Energy-Latency Attacks on Neural Networks
Figure 3 for Sponge Examples: Energy-Latency Attacks on Neural Networks
Figure 4 for Sponge Examples: Energy-Latency Attacks on Neural Networks

Share this with someone who'll enjoy it:

The high energy costs of neural network training and inference led to the use of acceleration hardware such as GPUs and TPUs. While this enabled us to train large-scale neural networks in datacenters and deploy them on edge devices, the focus so far is on average-case performance. In this work, we introduce a novel threat vector against neural networks whose energy consumption or decision latency are critical. We show how adversaries can exploit carefully crafted $\boldsymbol{sponge}~\boldsymbol{examples}$, which are inputs designed to maximise energy consumption and latency. We mount two variants of this attack on established vision and language models, increasing energy consumption by a factor of 10 to 200. Our attacks can also be used to delay decisions where a network has critical real-time performance, such as in perception for autonomous vehicles. We demonstrate the portability of our malicious inputs across CPUs and a variety of hardware accelerator chips including GPUs, and an ASIC simulator. We conclude by proposing a defense strategy which mitigates our attack by shifting the analysis of energy consumption in hardware from an average-case to a worst-case perspective.

View paper onarxiv icon

Share this with someone who'll enjoy it: